I was already expecting your e-mail! It never lates when some feature is announced/planned! ;-)
Ok. I'll answer briefly to your questions, but don't ask much more because I can't and won't tell ya! On 2002.04.18 23:45 Sergey V. Udaltsov wrote: > > At this moment I'm trying to debug to make things work barely before > I'll > > make a new branch: mach64-0-0-4-branch. But tommorrow I'll create it > Wow! So would you mind giving a shout here when something testable > appears on FTP? Of course. I'll announce when it when it works. Testing will be fundamental so that we don't get nasty hangs when we do switch to DMA. > So finally, did I get it right about this new branch?: > 1. A lot of code is moved into DRM module > 2. No real DMA yet. > 3. DMA emulation is slow. Yep. > How do you think - will 0-0-4 finally be really slower than 0-0-3? I've made a quick test of the DMA emulation on the Mesa driver and it was real slow (1:3). This could be improved (the emulation isn't done in parallel, .e.g), but there's no point in loosing time with that. When the rest is rock solid we just switch to real DMA and forget about it. > Also, will 0-0-4 someday have DMA - or there will be 0-0-5 for it? Don't think that there will be need of 0-0-5 as 0-0-4 will be pretty useless for real applications without it. > > Cheers, > > Sergey > Again I stress that testing this branch, although very slow, will be crucial, as DMA will be just at the corner. I believe that the amount of work required to be ready for starting the DMA work will roughly the same that was necessary to port the mach64 driver to Mesa 4.x. Regards, José Fonseca _______________________________________________ Dri-devel mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/dri-devel
